"Ban the tush push." Change the definition and check.
The data has no name for the play, so the tool measures only a written definition. Here you write it: who carried, how many yards to go, the formation, and the downs. Each combination is a different dictionary entry, and the page shows what the tool printed for it: the definition, every filter with the plays left after it, and how often the defined play converted against every other short-yardage rush. This is the same output the agent reads.
Loading the tool's answersThe bar is a scale of plays converted, from 0% at the left to 100% at the right, with a line every ten and a heavy line at 50%. The yellow marker is the conversion rate of the defined play, the blue marker the rate of the other short-yardage rushes, and the hatched stretch between them is the gap. The two colors are the lines a TV broadcast paints on the field: blue the line of scrimmage, yellow the line to gain.
The definition
The filters, in order
The two groups
| Group | Plays | Converted | Rate |
|---|
Not in this output: whether the play should be legal. A gap is how often each kind of play converted in the plays kept. It does not say why, and it does not say what the rule should be.
A downs choice here is part of the definition, so it narrows only the defined play. The comparison stays every other rush with the same yards to go, on any down. The tool's --downs flag, which the agent can add to a command, narrows both groups instead.
Who ran it, season by season
The first question about any play: whose was it? The same tool with --by team, one season at a time, for the reference definition. One row per team, one column per season, and the number is the plays that matched the definition. Each season's most is in yellow; the first row is every team together. The tool lists a team once it has 10 matching plays in a season, so a smaller count reads "under 10". The last column is the five seasons together, and 2026 is the season so far.
Remember what the data cannot see: who pushed. This is who ran the play the definition describes, a quarterback run with two or fewer yards to go, not from shotgun. Every team has that play. Change the definition above and the explorer measures it; this table does not change with the buttons.
The reference definition, by season and by team
The same tool with --by season and --by team, for the reference definition: a quarterback run with two or fewer yards to go, not from shotgun, on any down. A team gets a row once it has 10 or more defined plays.
Season by season, 2021 to 2025
| Season | Defined plays | Rate | Other rushes | Rate | Gap |
|---|
Team by team, 2021 to 2025
| Team | Defined plays | Rate | Other rushes | Rate | Gap |
|---|
Team by team, 2026
| Team | Defined plays | Rate | Other rushes | Rate | Gap |
|---|
The four searches by name
Before any definition, the search: how many play descriptions, 2021 through 2026, contain the words. A description matches when it contains them, in any case, anywhere in the text.
About these answers
Nothing here is computed in your browser. A script, build-explorer-data.mjs, wrote each of the 54 definitions these buttons can build as an entry in a scratch dictionary, asked the tool to measure each one for seven sets of seasons, 378 answers in all, saved what it printed, and removed the scratch dictionary. The page divides the counts the tool printed and rounds once, the same way the tool does.
Only the reference definition has a dictionary entry of its own: tush-push, which the agent writes on air on October 26, 2026, and which goes into the repo afterward. The other 53 were written to the scratch dictionary only. Every command on this page names the scratch entry, as the script ran it, so against the repo's dictionary the tool refuses it: no entry of that name is on file. For the reference definition, use --play tush-push once the episode's entry is in the repo. To measure any other, write it as an entry first.